The Interferon Type I/III Response to Respiratory Syncytial Virus Infection in Airway Epithelial Cells Can Be Attenuated or Amplified by Antiviral Treatment

ABSTRACT

Human respiratory syncytial virus (RSV) is a single-stranded RNA virus that causes acute, and occasionally fatal, lower respiratory illness in young infants, the elderly, and immunocompromised patients. Therapeutic interventions able to cut short viral replication and quickly return the airways to normal function are needed. An understanding of antiviral activities and their effects on host defense mechanisms is important for the design of safe and effective therapy. We targeted functionally and temporally distinct steps within the viral life cycle using small-molecule RSV inhibitors and studied their antiviral activities and their effects on innate interferon responses of airway epithelial cells in vitro. Antivirals acting upstream of RSV polymerase activity (i.e., compounds targeting the fusion protein or the nucleoprotein) reduced viral load immediately postinfection and partially attenuated interferon responses. In contrast, antivirals directed to the RSV polymerase demonstrated activity throughout the viral replication cycle and specifically modulated the RIG-I/mitochondrial antiviral signaling protein (MAVS)/TBK1/IRF3/interferon-stimulated gene (ISG) axis, causing either an upregulation or a downregulation of interferon responses, depending on the mechanism of polymerase inhibition. Notably, polymerase inhibition leading to the accumulation of abortive RNA products correlated with the amplification of interferon-stimulated genes to up to 10 times above normal infection levels. Understanding how antiviral activities and their modulation of innate immunity may affect recovery from RSV infection will help guide the development of safe and effective therapies.

IMPORTANCE RSV circulates seasonally, causing acute lower respiratory disease. Therapeutic interventions with efficacy throughout the viral replication cycle, rapid viral clearance, and prevention of potentially harmful inflammatory responses are desirable. Compounds targeting the RSV polymerase inhibited virus replication late in the viral life cycle and, depending on the functional domain targeted, either attenuated or amplified RIG-I and downstream interferon pathways in infected cells. These data will help guide the development of safe and effective therapies by providing new molecular evidence that the mechanism of inhibition by an antiviral compound can directly impact innate antiviral immune responses in the airway epithelium.

INTRODUCTION

Human respiratory syncytial virus (RSV) is a single-stranded, negative-sense RNA virus belonging to the family Paramyxoviridae. Premature and very young infants are at the highest risk of having severe lower respiratory tract infections and, later, a higher incidence of chronic asthma (1). Older adults with chronic lung or heart disease and individuals with suppressed immune systems also often require medical care after RSV infection. The lack of long-lasting protection after primary infection contributes to the capacity of RSV to cause yearly outbreaks and has challenged vaccine development (2). Synagis, a monoclonal antibody (MAb) directed against the RSV fusion protein, has shown prophylactic utility, but its cost and intramuscular route of administration have limited its use to hospitalized, high-risk infants <2 years of age (2). The only therapeutic intervention for RSV infection currently available to patients is the use of ribavirin (3), but its use is also limited because of poor efficacy and teratogenicity and the requirement of an aerosol and/or intravenous (i.v.) mode of administration in hospital settings (4). New therapeutic interventions able to lower the viral load, decrease transmission, and prevent lower respiratory complications are needed.

RSV infection is initiated by attachment of the viral G protein to the surface of airway epithelial (AE) cells (5, 6). Following attachment, the viral F protein mediates fusion of the viral and cellular membranes, allowing the RSV ribonucleic protein (RNP) complex, comprised of the viral RNA genome encapsulated with nucleoprotein (N) and associated with the phosphoprotein (P) and RNA-dependent RNA polymerase (L), to enter the cytoplasm. The RNP complex performs viral transcription to produce capped and polyadenylated mRNAs and genome replication. Detailed understanding of RSV biology is made difficult by the intimate coupling of transcription and replication activities. Furthermore, the RSV polymerase is large and complex, containing multiple domains and enzymatic activities that allow it to function and be regulated as a transcriptase (including mRNA capping activity) and replicase (6–8).

The dominant intracellular response to infection of the airway epithelia by RSV is the activation of innate antiviral type I and III interferon (IFN) signaling after detection of viral nucleic acids by pattern recognition receptors (9, 10, 11). Cytoplasmic viral RNA is detected by helicase receptors such as RIG-I. These receptors bind short RNA products, functionally couple with the mitochondrial antiviral signaling protein (MAVS), and activate downstream kinases and transcription factors, leading to interferon (IFN) production and upregulation of interferon-stimulated genes (11, 12).

In this study, we used small-molecule inhibitors targeting unique steps in the RSV replication cycle to examine how the mechanism of inhibition affects the innate immune response to the virus. Several direct-acting, small-molecule inhibitors targeting different steps of the RSV life cycle were tested, including TMC-353121, an entry inhibitor targeting the RSV fusion protein (13); RSV-604, an early postentry inhibitor binding a conserved region of the nucleocapsid protein (14, 15); YM-53403, an inhibitor targeting the RSV polymerase in variable region 2 (16); and BI-D, an inhibitor targeting the RSV polymerase in domain V (17). The mechanisms of inhibition by RSV-604 and YM-53403 are not yet understood, while BI-D has been shown to cause iterative, abortive transcription leading to an accumulation of short (<50-nucleotide [nt]) viral RNA products which are triphosphorylated but do not contain a 5′ cap (18). To investigate the interactions between RSV infection, innate immunity, and RSV inhibitors, RSV-A2 infection was established in cells a day before treatment with compounds, and the antiviral and cellular responses were examined 2 days postinfection (pi). Multiple antiviral assays were carried out to correlate the observed transcriptional and translational changes in host cell immunity with the mechanism of compound action. Here, we show that different inhibitors have significantly different impacts on the innate immune response which correlate with their effect on viral RNA quantity and quality. The in vivo relevance of the findings was supported by the use of organotypic cultures derived from the differentiation of primary human tracheal/bronchial epithelial cells at the air-liquid interface.

MATERIALS AND METHODS

Cell culture and viruses.

HEp-2 cells (CCL-23) and A549 cells (CCL-185) were obtained from American Type Culture Collection and cultured in minimal essential medium (MEM) (1×) and GlutaMAX with Earle's salts (Life Technologies) or F12K nutrient mixture (Kaighn's modification) with l-glutamine (Life Technologies), respectively, supplemented with 10% heat-inactivated fetal bovine serum (HI-FBS; HyClone), 10 units/ml penicillin, and 10 μg/ml streptomycin. The MT4 cell line was obtained from the NIH AIDS Research and Reference Reagent Program (Germantown, MD) and cultured in RPMI 1640 medium (Irvine Scientific) supplemented with 10% FBS, 100 units/ml penicillin, 100 units/ml streptomycin, and 2 mM l-glutamine. RSV replicon BHK cells (19) were grown in MEM (1×) containing GlutaMAX with Earle's salts (Life Technologies), 10% HI-FBS, 10 units/ml penicillin, 10 μg/ml streptomycin, and 50 μg/ml blasticidin S. Air-liquid-interface differentiated primary normal human airway epithelial (HAE) cells were obtained from MatTek Corporation (AIR-100 kit). The HAE cells were derived from trachea and mainstem bronchi of lung tissue donated with informed consent for research purposes at the time of death. Three healthy, nonsmoking, Caucasian adult donors (donor 1 [d1], a 23-year-old male [ID9831], d2, a 23-year-old male [ID11257], and d3, a 33-year-old female [ID11581]) were used. HAE cells were grown in Millicell CM single-well tissue culture plate insertions (Millipore) (pore size = 0.4 μm; inner diameter = 0.9 cm; surface area = 0.6 cm²) in 6-well dishes with defined Dulbecco's modified Eagle's medium (DMEM)-based media provided by MatTek. Directly pelleted 1,000× concentrated stocks of RSV-A2 (10^8.25 50% tissue culture infective doses [TCID₅₀]/ml) were purchased from Advanced Biotechnologies Inc.

Cell-based RSV ELISA.

A549, HEp-2, or BHK cells were plated at a density of 3,000 cells/well in black, flat-bottom, clear-bottom, 96-well plates (Corning) and cultured overnight in growth medium. Cells were infected with RSV-A2 at a multiplicity of infection (MOI) of 0.1 PFU/cell in 0.2 ml of growth media containing 3-fold serially diluted compounds. After 3 h, the inoculum was removed, the cells were washed once with 0.2 ml of growth media, and 0.2 ml of fresh growth media containing 3-fold serially diluted compound was then added and the reaction mixture incubated for 3 days. Following the incubation, RSV replication was quantified using a colorimetric anti-F protein cell-based enzyme-linked immunosorbent assay (ELISA). Medium was aspirated from the wells, and the cells were fixed with 100 μl of 4% paraformaldehyde–phosphate-buffered saline (PBS) for 10 min. Wells were washed 1× with PBS containing 0.05% Tween 20 (PBS-T; Anatrace) and blocked for 1 h using 100 μl of Superblock–PBS (Thermo Scientific). Blocking buffer was removed, and 50 μl of mouse anti-RSV F protein MAb 858-1 (Millipore) diluted 1:1,000 in Superblock was added with gentle agitation at ambient temperature for 2 h. Following three washes with 0.2 ml PBS-T, 50 μl of a goat anti-mouse IgG horseradish peroxidase (HRP) secondary antibody (Sigma), diluted 1:2,000 in Superblock, was added with gentle agitation at ambient temperature for 1 h. After three washes with 0.2 ml PBS-T, 25 μl of TMB supersensitive substrate (Sigma) was added and the signal read at 450 nm on a VERSAmax reader (Molecular Devices).

In vitro cytotoxicity.

A549 or HEp-2 cells were plated in 96-well plates at a density of 3,000 or 10,000 cells per well and allowed to attach overnight at 37°C. Following attachment, the medium was replaced with 200 μl of fresh medium containing 3-fold serially diluted compound with a concentration ranging from 15 nM to 100 μM. Cells were cultured for 4 days at 37°C. Following the incubation, the cells were allowed to equilibrate to 25°C and cell viability was determined by removing 100 μl of the cell culture medium and adding 100 μl of CellTiter-Glo viability reagent (Promega). The mixture was incubated for 10 min at 25°C, and the luminescence signal was quantified on a Victor luminescence plate reader (Perkin-Elmer).

A high-throughput 50% cytotoxic concentration (CC₅₀) assay was used for MT4 cytotoxicity evaluation. Complete RPMI 1640 media containing 100× concentrations of 3-fold serially diluted compound, ranging from 26 nM to 530 μM, were stamped in quadruplicate into black 384-well plates. After compound addition, 2 × 10³ MT4 cells were added to each well using a MicroFlo liquid dispenser (BioTek) and the cells were cultured for 5 days at 37°C. Following the incubation, the cells were allowed to equilibrate to 25°C and cell viability was determined by adding 25 μl of CellTiter-Glo viability reagent, incubating the mixture for 10 min at 25°C, and reading the luminescence signal on a Victor luminescence plate reader.

Reverse transcription-quantitative PCR (RT-qPCR).

The N gene probe and primers were ordered from IDT at a 2:1 primer/probe ratio (primer 1, GCTAGTGTGCAAGCAGAAATG; primer 2, TGGAGAAGTGAGGAAATTGAGTC; double-quenched probe, FAM/ATTGGGTGG/ZEN/AGAAGCAGGGTTCTAC/IABFQ) (FAM, 6-carboxyfluorescein; ZEN, ZEN internal quencher; IABFQ, Iowa Black FQ quencher).

A human GAPDH (glyceraldehyde-3-phosphate dehydrogenase) endogenous control (FAM/MGB probe; 4352934E) was ordered from Applied Biosystems (ABI, Life Technologies). Single-tube TaqMan gene expression assay primers were used to independently confirm transcriptome sequencing (RNA-seq) data. Primer/probe FAM-MGB stocks (4331182) (20×) were ordered from ABI for assays of innate immunity genes, including RIG-I/DDX58 (Hs00204833_m1), MDA-5/IFIH1 (Hs01070332_m1), IRF3 (Hs01547283_m1), IFN beta 1 (IFN-β1; Hs01077958_s1), MX2 (Hs01550811_m1), and NF-κB (Hs00765730_m1). Total cellular RNA was purified using an SV total RNA isolation kit (Promega). A TaqMan RNA-to-CT 1-step kit (ABI) was used for RT-PCR amplification and quantification, using a 7900HT Fast Real Time PCR System (Applied Biosystems), under the following conditions: 15 min of reverse transcription at 48°C and 10 min of activation at 95°C followed by 40 cycles of denaturation at 95°C for 15 s and annealing and extension at 60°C for 1 min.

Transcriptome analysis.

A549 cells were plated in 6-well plates (Costar Falcon) at a density of 2 ×10⁵ cells/well and allowed to attach overnight. Following attachment, the cells were either mock infected (media alone) or infected with RSV-A2 at an MOI of 0.5. After 3 h of incubation at 37°C, the virus inoculum was removed and the cell monolayer washed once with 2 ml of 5% HI-FBS media. Following the wash, 2 ml of 5% HI-FBS medium was added and the infection was allowed to continue for 24 h at 37°C. At 24 h postinfection, the medium was replaced with 2 ml of fresh 5% HI-FBS media containing either dimethyl sulfoxide (DMSO) or compounds at a concentration 20-fold over their respective 50% effective concentration (EC₅₀) values. Forty-eight hours postinfection, the cells were washed once with PBS, scraped into 1 ml of PBS, and centrifuged at 5,000 × g for 5 min. Cell pellets were frozen at −70°C and shipped to Expression Analysis, Inc., for RNA purification, library synthesis [TruSeq mRNA kit; 500 ng poly(A) RNA input], and paired-end Illumina HiSeq 2500 sequencing at a depth of 25 million reads. Transcriptome reads across the RSV genome were mapped and normalized to the total number of mapped transcriptome reads in the same sample (data not shown). Host cell gene transcript numbers were calculated in both infected and mock-infected samples as the fold change relative to the uninfected DMSO sample level (genes with values of zero were assigned a value of 1). After normalization was performed by subtracting the values corresponding to the fold mock infection changes for each compound, data were analyzed using the MetaCore pathway analysis database (GeneGo; Thompson Reuters) and a filter of 5-fold significance. Quantitative comparisons of gene expression are complicated by the half-lives and feedback loops affecting each mRNA. Also, the experimental 24-h treatment delay differently affects each antiviral compound's efficacy (particularly those of TMC-353121 and RSV-604 versus YM-53403 and BI-D); therefore, differences in residual viral loads should be evaluated with caution.

ELISA analysis of cytokines.

Cytokines in supernatants from A549 cells or in the basal media of HAE cells, treated in the same way as described for the transcriptome analysis, were measured in the VeriKine human IFN-β ELISA (PBL Assay Science), the Proteintech human interleukin-28B (IL-28B) ELISA, the Sigma-Aldrich human IL-28A and IL-29 ELISAs, and the Life Technologies 25-Plex human cytokine assay. Supernatants contained 0.5% Triton X-100 to neutralize virus. Control treatments with the compounds in mock infections were performed to rule out off-target or nonspecific activation of cytokines.

Western blot analysis of RSV and RIG-I pathway.

A549 cell pellets were lysed on ice for 10 min in PBS [pH 7.4] containing 1% NP-40, 10% glycerol, and a 1:100 dilution of protease/phosphatase inhibitors (Cell Signaling Tech). Cell lysates were clarified by centrifugation at 10,000 × g for 10 min, and the protein concentration was determined using a Pierce bicinchoninic acid (BCA) protein assay kit. Lysate protein (50 to 80 μg) was loaded onto a 4% to 12% Bis-Tris polyacrylamide gel (Life Technologies) and electrophoresed using morpholineethanesulfonic acid (MES) SDS running buffer (Life Technologies). Proteins were transferred to polyvinylidene difluoride (PVDF) using an iBlot system (Life Technologies). A RIG-I pathway antibody sampler kit containing rabbit MAb directed against human RIG-I, MDA-5, MAVS, IκB kinase ε (IKKε), phospho-IKKε, TBK1, phospho-TBK1, IRF3 and phospho-IRF3, rabbit MAb phospho–NF-κB p65, and rabbit MAb NF-κB p65 was purchased from Cell Signaling Technology and used following the supplied instructions. Goat polyclonal antibody (pAb) anti-respiratory syncytial virus antibody (ab20531; Abcam) was used at a dilution of 1:800 for 2 h at ambient temperature in PBS-T containing 5% nonfat milk (Bio-Rad). Following the incubation, the membrane was washed 3 times with PBS-T and a rabbit anti-goat IgG H&L-HRP secondary antibody (Abcam) diluted 1:5,000 in PBS-T containing 5% nonfat milk was added for 1 h at ambient temperature. Following the incubation, the membrane was washed 3 times with PBS-T. All blots were developed using SignalFire Plus ECL reagent (Cell Signaling Technology). Control treatments with the compounds in mock infections were performed to rule out off-target or nonspecific effects on protein expression.

Time of compound addition.

HEp-2 or A549 cells were seeded at a density of 10,000 cells/well in black, clear flat-bottom 96-well plates (Corning). Cells were infected with RSV-A2 at an MOI of 1 PFU/cell in growth media. To synchronize the RSV infection, the virus was first allowed to bind to the cell surface at 4°C for 4 h and then the culture was shifted to 37°C. After 1 h, virus entry was terminated by removal of the inoculum and the addition of a 200 nM concentration of an RSV fusion inhibitor, GS-5806, prepared in house (20). RSV inhibitors were added at selected time points, and RSV transcription was quantified using qPCR of the RSV N gene at 36 h postinfection. All data were normalized to cellular GAPDH. A 20-fold excess over the EC₅₀ of each compound was used to normalize for potency differences and to ensure that efficacy was not reflected in time-of-addition (ToA) profiles. For treatment window experiments using multicycle infections of HAE cells, the transwells were washed in PBS to remove mucus and infected apically for 3 h with a 1:600 dilution of RSV-A2. Excess virus was washed off with PBS, and 25-fold EC₅₀ concentrations of inhibitors were added to the basal media at 0, 24, and 48 h. All wells were harvested at 72 h postinfection and their contents quantified relative to those of DMSO controls by RT-qPCR analysis of the RSV N and GAPDH genes.

RSV replicon.

BHK cells containing a stably transformed subgenomic RSV replicon expressing a green fluorescent protein (GFP) reporter cassette were licensed from Apath, LLC (19). Replicon cells were seeded at a density of 10,000 cells/well in black, flat, clear-bottom 96-well plates (Corning) and cultured overnight in growth media containing 50 μg/ml blasticidin S. The growth medium was aspirated and replaced with 200 μl of 3-fold serially diluted compounds prepared in 2% HI-FBS media containing 50 μg/ml blasticidin S. After 3 days at 37°C, the cells were fixed using 4% paraformaldehyde (PFA) supplemented with Hoechst dye (Invitrogen). Replicon gene expression and cytotoxicity were simultaneously quantified on a Cellomics high-content imaging system (Thermo Scientific) by measuring the average replicon GFP signal and cell number, respectively.

RSV RNP assay.

RSV ribonucleic protein (RNP) complexes were isolated from RSV-infected HEp-2 cells through biochemical fractionation as described previously (17). Transcription reaction mixtures contained 25 μg of crude RSV RNP complexes in 30 μl of reaction buffer {50 mM TRIS-acetate (pH 8.0), 120 mM potassium acetate, 5% glycerol, 4.5 mM MgCl₂, 3 mM dithiothreitol (DTT), 2 mM EGTA [ethyleneglycol-bis(2-aminoethylether)-tetraacetic acid], 50 μg/ml bovine serum albumin (BSA), 2.5 U RNasin (Promega), ATP, GTP, UTP, CTP, and 1.5 uCi [α-³²P]CTP [3,000 Ci/mmol stock supplied at 10 μCi/μl in 50 mM Tricine (pH 7.6; PerkinElmer)]}. Cold, competitive CTP was added at a final concentration of 2 μM (half of its measured K_m value). The three remaining nucleotides were added at a final concentration of 100 μM. RSV inhibitors were prepared at 300 μM concentrations, and 5 μl was diluted in 25 μl of the RNP mixture, with DMSO normalized. Following incubation at 31°C for 1.5 h, 420 μl of RLT buffer was added and the newly transcribed RNP transcripts were purified on RNeasy mini columns (Qiagen). RNA was eluted in 2× 30 μl of water, and 20-μl aliquots were prepared for analysis in 10 μl of RNA sample loading buffer (Sigma). After being heated at 65°C for 10 min and cooled on ice, RNA was separated using 2 M formaldehyde–1.2% agarose gel electrophoresis in MOPS (morpholinepropanesulfonic acid) buffer containing 40 mM MOPS, 10 mM sodium acetate, and 1 mM EDTA. The agarose gel was dried and exposed for 5 days to a Storm phosphorimager screen and developed using a Storm phosphorimager (GE Healthcare).

Recombinant RSV L/P assay.

RNA was synthesized in vitro using a purified complex of RSV-A2 L and P (L/P) proteins, produced from recombinant baculovirus in insect cells as described previously (21). The RNA synthesis reactions were performed in a 50-μl reaction volume that contained 2 μM RNA oligonucleotide template, consisting of nucleotides 1 to 25 of the RSV trailer complement promoter sequence, in a buffer containing 50 mM Tris-HCl (pH 7.4), 8 mM MgCl₂, 5 mM DTT, 10% glycerol, 1 mM (each) recombinant ATP (rATP), rCTP, rGTP, and rUTP, and 10 μCi of [α-³²P]ATP (PerkinElmer) (3,000 Ci/mmol stock supplied at 10 μCi/μl in 50 mM Tricine [pH 7.6]). Each RSV inhibitor compound was dissolved in DMSO and included in the reaction at a final concentration of 50 μM. The reaction mixtures were incubated at 30°C for 5 min, and then 1 μl of wild-type (wt) L/P complex was added to each reaction mixture. The positive-control reaction mixture contained equivalent volumes of DMSO and wt L/P complexes. Negative-control reaction mixtures contained an equivalent volume of DMSO, and either the L/P complex was omitted or a mutant variant of L containing an alanine substitution at amino acid N812 was used. Following the incubation, reaction mixtures were incubated at 90°C for 3 min to inactivate the polymerase and 50 μl of stop buffer (20 mM EDTA–deionized formamide containing xylene cyanol and bromophenol blue) was added. Twenty microliters of each sample was analyzed by denaturing gel electrophoresis on a 7 M urea–20% acrylamide gel alongside RNA ladders that represented the products from the +3 and +1 sites (ladders 1 and 2, respectively).

RESULTS

Effect of antiviral compounds on RSV RNA and protein accumulation in cell culture.

The goal of this study was to examine the innate immune response to RSV infection in human A549 epithelial cells treated with RSV inhibitors TMC-353121, RSV-604, YM-53403, and BI-D (Table 1). First, the EC₅₀ of each compound was determined using a colorimetric, cell-based ELISA quantifying the expression of the RSV F protein. Cells were infected with RSV at a low MOI and incubated for 3 days, with each compound present at the time of infection and until the cells were harvested. The EC₅₀s, shown in Fig. 1A, and cytotoxicity evaluations (Table 1) were used to inform subsequent experiments.

TABLE 1.

RSV inhibitor antiviral potency and cytotoxicity in cell lines^a

^aInfectivity and cytotoxicity values (± standard deviations) are averages of the results of a minimum of 3 independent experiments.

FIG 1.

Antiviral activities of RSV inhibitors in A549 cells. (A) The EC₅₀ concentration of each compound was determined in a colorimetric RSV F protein cell-based ELISA. The x axis data show the concentrations of compound, and the y axis data represent the measured levels of infection relative to a DMSO control. (B) A 20-fold excess of compound EC₅₀ concentrations was used to treat A549 cells at 24 h post-RSV infection. Cells were harvested at 48 h postinfection and the relative levels of viral RNA determined by sequencing 500 ng of total cellular poly(A) RNA (RSV transcript numbers are indicated above the bars). The data in panel B were verified by RT-qPCR assays (n = 3). (C) Cells were treated and harvested as described for panel B and the levels of viral protein detected in 50 μg total protein/lane using pAb anti-RSV Western blot analysis (lane 1, mock infection; lanes 2, 3, 4, 5, and 6, RSV infected with DMSO, TMC-353121, RSV-604, YM-53403, and BI-D, respectively). The data shown in panels A and C are representative of the results of 3 independent assays.

To understand the effects of inhibitors on the cellular antiviral response to RSV, we examined how much RSV RNA and protein expression there was when the inhibitors were added after infection, which would enable the antiviral response to be initiated. A549 cells were infected with RSV at an MOI of 0.5 and incubated for 24 h, at which time the inhibitors were added at concentrations 20-fold over their EC₅₀s. The cells were incubated an additional 24 h and then harvested. Under these conditions, syncytia were visibly present in all treatment groups, and, in the BI-D-treated cells only, the majority of syncytia progressed to lysis. Accumulation of viral RNA was quantified relative to total RNA levels using RT-qPCR (Table 2) or RNA-seq (Fig. 1B), and the accumulation was expressed as a percentage of that seen with the DMSO-treated infection. Viral protein was analyzed by loading equal amounts of total cellular protein on SDS-PAGE followed by Western blotting performed with a polyclonal antibody to RSV (Fig. 1C). Not surprisingly, when used 24 h postinfection, the TMC-353121 fusion inhibitor reduced viral RNA and protein levels to about half of the level of the uninhibited DMSO control. RSV-604 was also only about 50% effective. In contrast, YM-53403 reduced accumulation of viral RNA by 80% and significantly reduced viral protein expression, while BI-D showed almost complete inhibition of expression of both viral RNA and protein (Western blotting of the same cellular lysates for host cell proteins did not show reductions; Fig. 2).

TABLE 2.

Percent change of the RIG-I-like interferon antiviral pathway upon RSV-A2 infection and treatment with antiviral compounds in A549 and primary HAE cells determined by RT-qPCR

Gene	Cell linea	% change of the RIG-I-like interferon antiviral pathway
		TMC-353121	RSV-604	YM-53403	BI-D
RSV Nb	A549	41 ± 4	44 ± 8	11 ± 2	4 ± 0.4
	HAE donor 1	48 ± 9	1.6 ± 0.1	18 ± 2	14 ± 10
	HAE donor 2	62 ± 6	27 ± 12	34 ± 10	34 ± 11
RIG-I	A549	103 ± 26	61 ± 28	39 ± 2	183 ± 49
	HAE donor 1	103 ± 8	35 ± 5	82 ± 4	639 ± 103
	HAE donor 2	135 ± 36	162 ± 34	154 ± 16	924 ± 50
IFN-β1	A549	30 ± 10	14 ± 4	9 ± 2	435 ± 123
	HAE donor 1	292 ± 24	205 ± 45	273 ± 11	741 ± 232
	HAE donor 2	88 ± 3	89 ± 29	103 ± 10	286 ± 13
MX2	A549	188 ± 20	84 ± 21	59 ± 7	207 ± 68
	HAE donor 1	59 ± 7	12 ± 2	36 ± 14	425 ± 40
	HAE donor 2	65 ± 25	81 ± 17	74 ± 6	596 ± 28

^aHAE cell analysis was performed with 3 donors, with each experiment performed in triplicate. The infection of the third donor was insufficient to provoke a response. A549 data were averaged from the results of 2 independent experiments.

^bGene data are expressed as a percentage of the results determined for each donor or untreated A549 cell DMSO infection (assigned a value of 100%)

FIG 2.

Helicase sensing of RSV-2 infection and polymerase inhibition in A549 cells. A549 cells infected for 24 h before treatment and harvested at 48 h postinfection were subjected to Western blot analysis of proteins in the innate viral RNA sensing pathway. Total protein was assayed and 50 μg/lane loaded for mock infection (lane 1) and RSV-A2-infected DMSO (lane 2), YM-53403 (lane 3), and BI-D (lane 4). RIG-I and MDA-5 were expressed only upon infection (top row blots, lane 1 versus lanes 2 to 4), whereas MAVS (top row, rightmost blot, lanes 1 to 4), NF-κβ p65, IKKε, TBK1, and IRF3 (middle row blots, lanes 1 to 4) were all constitutively expressed and unaffected by infection. Increased levels of phosphorylation of NF-κβ p65, TBK1, and IRF3 occurred upon infection (bottom blots, lane 1 versus lane 2). Levels of phosphorylation of TBK1 and IRF3 were differentially affected by the polymerase inhibitors, being decreased byYM-53403 (lane 2 versus lane 3) and unchanged by BI-D (lane 2 versus lane 4). These changes were observed in 3 independent experiments.

The activity of these inhibitors was further tested in differentiated cultures of primary human airway epithelial (HAE) cells. Levels of viral RNA from three donors were quantified relative to the level seen with a DMSO infection using the same 24-h-infection-24-h-treatment procedure by RT-qPCR of the RSV N gene and GAPDH (Table 2). One donor was not infected sufficiently to allow measurement of changes and was not included in the assays. The relative levels of RSV inhibition in two donors in the fusion and postfusion compound classes were variable but were consistent with the A549 cell data. However, the three postentry inhibitors were poorly discriminated. The relative exposures of infected cells in the HAE environment to these compounds could be limited by their different physicochemical properties.

Effect of antiviral compounds on the transcriptional response to RSV infection.

The dominance of the innate interferon immune response to RSV infection in the human airway is well characterized (9, 10, 11) and includes transcripts for RIG-I, MDA5, IFN-β1, CCL5, IFIT2, OASL, MX1, RSAD2, and many other interferon-stimulated genes (ISG). We infected A549 cells with RSV-A2, treated the cells with antiviral inhibitors from 24 h postinfection (pi) onwards, and then harvested the cells at 48 h pi (this corresponds to the experimental design used to obtain the data shown in Fig. 1B and C). Our analysis of host cell transcript profiles confirmed that the most significant pathway affected by RSV infection and antiviral treatment was the interferon type I response to viral RNA. Transcript counts show a partial downregulation of antiviral pathways in cells treated with three of the inhibitors, with YM-53403 treatment showing the greatest reduction followed by RSV-604 treatment and TMC-353131 treatment. In contrast, BI-D treatment led to an upregulation of expression of these genes to levels well above those measured in control infected cells treated with DMSO (Tables 2 and 3). No changes in these genes were observed in the corresponding mock-infected cells in the presence of antiviral compounds. The contrast in the antiviral responses revealed by the RIG-I-like sensors of viral RNA was best demonstrated by YM-53403 and BI-D treatment. For example, treatment of RSV-infected cells with YM-53403 led to 63%, 14%, 25%, and 59% of the normal infected levels of RIG-I, IFN-β1, RSAD2, and CCL5 transcripts, respectively. Under the same conditions, BI-D treatment led to 449%, 1,348%, 952%, and 794% increases in the transcription of these genes. This provides evidence that, despite both compounds demonstrating effective inhibition of viral gene expression (e.g., 81% for YM-53403 and 96% for BI-D; Fig. 1B), polymerase inhibitors can have differential effects on innate immune signaling. BI-D treatment of infected cells also led to broader increases in related intracellular immune responses (e.g., major histocompatibility complex [MHC] class I genes) and responses outside the RIG-I like pathways (e.g., IFN-α and IFN-γ). Transcript levels of some genes involved in the RIG-I signaling pathway such as those encoding MAVS, TBK1, and IRF3 remained constant upon infection or treatment (Table 3), suggesting a role for translational or posttranslational regulation.

TABLE 3.

Fold changes in antiviral pathway mRNA transcript counts upon RSV-A2 infection and treatment with antiviral compounds in A549 cells^a

Product category and gene	Fold change in antiviral pathway mRNA transcript counts
	DMSO	TMC-353121	RSV-604	YM-53403	BI-D
Pattern recognition receptors
DDX58 (RIG-I)	37	51	33	23	166
IFIH1 (MDA-5)	155	191	120	87	724
DHX58 (LGP2)	101	82	45	63	349
TLR3	12	30	20	11	48
TLR4	1	−2	−2	3	5
TLR7	1	2	1	−1	2
TLR9	2	2	2	1	1
Mitochondrial signaling
MAVS	−1	−1	1	−1	−1
TMEM173 (Sting)	10	8	2	6	17
TRAP	−2	−1	−2	−1	−2
TBK1	−1	1	−1	−1	1
IKBKG (Nemo)	−1	1	1	1	1
Transcription factors
IRF1	11	10	6	6	21
IRF3	−1	1	−1	1	1
IRF7	27	62	57	27	60
IRF9	24	33	32	23	53
NFKB1	2	2	2	2	3
NFKB2	5	4	4	3	7
Type I, II, and III interferons
IFN-α1	1	1	1	1	31
IFN-α10	1	1	1	1	10
IFN-α13	6	1	1	1	42
IFN-α7	1	1	1	1	16
IFN-β1	130	42	25	18	1,753
IL-6	508	159	66	55	954
IFN-γ	1	1	1	1	13
IL28A	383	239	48	74	2,161
IL28B	202	119	23	43	954
IL-29	1,778	647	265	259	6,013
ISGs
CCL5 (RANTES)	147	67	51	87	1,168
CX3CL1 (Fractalkine)	16	18	11	5	54
CXCL10 (IP-10)	253	196	62	97	3,069
CXCL11	321	187	86	145	2,357
IFI27	443	855	671	551	1,349
IFI44	1,695	2,703	1,469	1,127	4,487
IFIT2	387	355	232	217	3,005
IFIT3	159	220	118	116	761
IFITM1	345	862	258	418	1,312
ISG15	142	310	277	193	445
ISG20	19	19	17	10	54
OAS1	9	19	12	9	33
OAS2	1,041	2,282	1,075	988	3,183
OAS3	8	15	9	8	19
OASL	754	713	602	562	4,339
MX1	257	650	302	234	861
MX2	102	337	130	84	415
RSAD2 (Viperin)	228	211	140	57	2,170
TSLP	4	1	1	1	16

^aData are from the results of one RNA sequencing experiment. Selected genes were validated by RT-qPCR (Table 2) and protein expression as described in Results. ISGs, interferon-stimulated genes.

Consistent with our findings in A549 cells, we observed a strong amplification of RIG-1, IFN-β1, and MX2 transcriptional immune activation in HAE cells (relative to each donor's DMSO response to RSV infection) upon treatment with the BI-D compound (Table 2, last column). Attenuation of antiviral transcriptional responses in HAE cells treated with TMC-353121, RSV-604, or YM-53403 was weak and variable, and no decreases in IFN-β1 transcription levels were observed. The same trend of significant BI-D stimulation and no attenuation was also observed in the measurements of the levels of secreted cytokines (see section below and Table 4). The organotypic HAE cultures contain multiple cell populations, including both basal and apical epithelial cells, club cells, and goblet cells. Since only about 20% of the cells in the organotypic cultures are the ciliated cells (22) able to be infected by RSV, and since their response to RSV is weak compared to that of A549 cells, there may not be enough range in these assays (at 48 h postinfection) to measure antiviral attenuation.

TABLE 4.

Levels of secreted cytokines upon RSV-A2 infection and treatment with antiviral compounds in A549 or HAE cells^a

Cytokine	Level (pg/ml) under indicated conditions
	DMSO mock treatment			DMSO			TMC-353121			RSV-604			YM-53403			BI-D
	A549	HAE d1	HAE d2	A549	HAE d1	HAE d2	A549	HAE d1	HAE d2	A549	HAE d1	HAE d2	A549	HAE d1	HAE d2	A549	HAE d1	HAE d2
IFN-β1	OOR<	OOR<	OOR<	345 ±140	OOR<	OOR<	187 ±62	OOR<	OOR<	24 ±27	OOR<	OOR<	27 ±10	OOR<	OOR<	899 ± 392	OOR<	OOR<
IL-28A	OOR<	OOR<	OOR<	4,047 ± 1,022	OOR<	OOR<	1,473 ±450	OOR<	OOR<	195 ± 45	OOR<	OOR<	341 ± 29	OOR<	OOR<	>10,000	104 ±49	67 ±16
IL-29	OOR<	OOR<	OOR<	>10,000	OOR<	OOR<	5,533 ± 2,713	OOR<	OOR<	980 ± 124	OOR<	OOR<	1,627 ± 449	OOR<	OOR<	>10,000	354 ± 251	53 ±19
IFN-α	OOR<	OOR<	OOR<	28 ±7	OOR<	OOR<	33 ±7	OOR<	OOR<	OOR<	OOR<	OOR<	10 ±9	OOR<	OOR<	20 ±2	18 ±8	OOR<
IL-6	OOR<	28 ±6	17 ±2	590 ±74	29 ±9	25 ±8	253 ±12	18 ±3	14 ±8	112 ±9	7 ±0	21 ±3	74 ±6	18 ±9	19 ±2	297 ±21	31 ±10	22 ±16
IL-8	82 ±6	8,022 ±1,034	5,500 ±337	4,025 ±912	6,833 ± 1,180	7,906 ±859	2,641 ±964	4,070 ± 946	5,933 ± 2,029	1,193 ± 39	3,862 ± 194	8,738 ± 20	873 ± 188	4,027 ± 47	6,743 ± 1,646	2,327 ± 454	8,449 ± 389	7,309 ± 3,874
IL-15	OOR<	OOR<	OOR<	90 ±0	OOR<	OOR<	93 ±5	OOR<	OOR<	OOR<	OOR<	OOR<	9 ±7	OOR<	OOR<	244 ±11	274 ±17	161 ±31
IP-10	OOR<	24 ±12	22 ±7	127 ±2	61 ±35	50 ±3	112 ±1	65 ±1	44 ±17	17 ±4	24 ±10	38 ±5	50 ±3	44 ±7	49 ±38	318 ±10	3,383 ± 1,926	645 ±84
MIP1α	OOR<	OOR<	OOR<	194 ±116	OOR<	OOR<	52 ±27	OOR<	OOR<	6 ±1	OOR<	OOR<	9 ±3	OOR<	OOR<	30 ±18	18 ±6	9 ±2
MIP1β	OOR<	OOR<	OOR<	34 ±3	7 ±1	7 ±1	12 ±1	OOR<	OOR<	OOR<	OOR<	6 ±1	OOR<	OOR<	OOR<	9 ±4	10 ±2	6 ±4
MCP-1	699 ± 100	183 ±38	154 ±13	2,881 ±379	204 ±18	202 ±27	2,904 ± 668	157 ±9	157 ±3	747 ± 20	120 ±6	215 ±11	1,843 ± 115	152 ±15	173 ±9	1,546 ± 284	262 ±21	184 ±33
RANTES	OOR<	OOR<	OOR<	1,621 ±22	OOR<	OOR<	1,757 ± 345	OOR<	OOR<	774 ± 177	OOR<	OOR<	1,425 ± 161	OOR<	OOR<	1,671 ± 43	22 ±7	11 ±1

^aIFN-β1, IL-28A, and IL-29 levels were measured by ELISA (OOR< [below detectable range], 50 pg/ml); the levels of the other cytokines were measured by Luminex assay (OOR<, 5 to 20 pg/ml). IL-28B, granulocyte-macrophage colony-stimulating factor (GM-CSF), eotaxin, MIG, tumor necrosis factor alpha [TNF-α[rsqb], IL-1β, IL-2, IL-4, IL-5, IL-7, IL-10, IL-12, IL-13, and IL-17 were not detected. A549 data represent the averaged results of three independent experiments. HAE assays were performed in triplicate for each of the 3 donors (d1, d2, and d3); the third donor's infection was insufficient to provoke a cytokine response.

Effect of RSV polymerase inhibitors on RIG-I-mediated protein responses to RSV infection.

The opposing effects of the YM-53403 and BI-D RSV polymerase inhibitors on the attenuation and amplification of RIG-I transcriptional pathways, respectively, were further investigated at the translational level. The net effects of infection and treatment on key proteins involved in the interferon response pathway were compared using ELISAs and Western blotting. In A549 cells, the RIG-I, MDA-5, and type I and III interferon proteins were undetectable until infection (Fig. 2 and Table 4), while MAVS and downstream transcription factors were constitutively expressed, and their levels of expression did not change upon infection (Fig. 2). No significant differences in the levels of RIG-I, MDA-5, MAVS, or selected transcription factors were detected upon treatment of infected cells with either polymerase inhibitor. However, Western blotting with anti-phosphoserine antibodies showed that the TBK1, IRF3, and NF-κB p65 transcription factors were activated by infection and differentially modulated by the two RSV polymerase inhibitors (Fig. 2, bottom panel). Treatment of RSV-infected cells with YM-53403 resulted in reduced phosphorylation of TBK1 and IRF3, and attenuation of these pathways resulted in an approximately 10-fold reduction of levels of secreted interferons compared to the results seen with untreated, infected cells (IFN-β1, IL-28A, and IL-29; Table 4). In contrast, treatment with the BI-D compound did not decrease the activation of TBK1 or IRF3, and the levels of secreted interferon increased ∼3-fold over the control infection levels (Table 4). This increase in the interferon levels seen with BI-D occurred despite the fact that at that time, the cells contained only 4% of the level of viral RNA seen with control infections (Fig. 1B). Secreted type I/III interferons are known to act in an autocrine and paracrine manner to transcriptionally activate ISG such as Toll-like receptor 3 (TLR3), ISG15, OAS2, MX1, MX2, IP-10, and Viperin, and these genes were upregulated in BI-D- but not YM-53403-treated infections (Table 3). No changes in the activation of NF-κB p65 were observed with treatment by either compound (Fig. 2, bottom panel), consistent with the regulation of this transcription factor by a distinct pathway independent of replication (23).

We found the cytokine response in HAE cells to RSV infection to be more restricted than that in A549 cells and too weak to allow measurement of attenuation. However, we were able to measure the specific amplification of several cytokines upon infection and BI-D treatment in a manner consistent with the observations in A549 cells. Among 29 cytokines measured, IL-6, IL-8, IP-10, and monocyte chemoattractant protein 1 (MCP-1) were detected in the basal media of the HAE cell cultures; after RSV infection, IP-10 levels increased (∼2-fold) and expression of IL-15 and RANTES was significantly induced (Table 4). Upon infection and treatment with the BI-D inhibitor, levels of IP-10, IL-15, and RANTES showed significant increases that were on the order of 10- to 100-fold, and type III interferons IL-28A and IL-29 could be detected. These changes were not observed in mock infections treated with BI-D. In contrast to the results seen with the A549 cells, we were unable to measure any IFN-β1 secreted into the basal (or apical) HAE compartments.

ToA profiles of RSV inhibitors.

The data described above show that treatment of RSV-infected cells with different RSV polymerase inhibitors resulted in very different innate responses to the infection. As none of the inhibitors had an effect on innate immune responses in mock-infected cells, the observed differences presumably reflect differences in the levels of RSV RNA expression in the treated cells that occurred as a consequence of the presence of the inhibitors. Although these inhibitors were previously characterized, they have not been compared directly with each other in the same assays, and the postentry mechanisms are not fully defined. Therefore, we compared the small-molecule inhibitor activities in a variety of assays to investigate how the antiviral kinetics and mechanisms of action correlate with the diverse effects on the activation of innate immunity that we observed. A time-of-addition (ToA) experiment can reveal mechanistic differences between classes of inhibitors by measuring how long compound addition can be postponed before antiviral activity is lost during a single infectious cycle (24). Three unique ToA profiles were observed in both the A549 and HEp-2 cell lines and are shown in Fig. 3. As expected for a fusion inhibitor, TMC-353121 demonstrated a sharp loss of activity when addition of compound was delayed past the first hour of infection. The RSV-604 N-protein inhibitor gradually lost 100% of its activity over time in HEp-2 cells and lost up to a maximum of 25% of its activity in A549 cells. This finding is consistent with activity occurring at a postentry step but prior to (or during) viral replication complex formation. YM-53403 and BI-D represented a third profile of activity by inhibiting RSV when compound was added at any time postinfection in both cell lines, consistent with both compounds being inhibitors of polymerase activity rather than of assembly of the replication complex. The infectivity of virus progeny produced after infection in A549 cells in the presence of 10 μM RSV-604, 2 μM YM-53403, or 0.1 μM BI-D, added 12 h postinfection and measured by plaque assay in HEp-2 cells, suggested the absence of significant effects of these compounds on progeny virus infectivity or egress in these cell lines (not shown).

FIG 3.

Compound time-of-addition (ToA) profiles for RSV inhibitors. (A) Single-cycle RSV-A2 infection of A549 (dashed lines) or HEp-2 (solid lines) cells. TMC-353121, RSV-604, YM-53403, or BI-D and ribavirin (80 μM) were added at −4, 0, 2, 4, 6, 8, 12, 16, 20, and 24 h postinfection as indicated on the x axis. All cells were harvested at 36 h postinfection, and the relative amounts of RSV N-gene/GAPDH were determined by RT-qPCR and are expressed as a percentage relative to the averaged data from DMSO-treated infections (n = 6). (B) Multicycle RSV-A2 infections of HAE cells (donor 1) were treated at 0, 24, and 48 h postinfection with compounds as described for panel A. Cells were harvested at 72 h postinfection and analyzed as described for panel A. Data shown represent averages of the results of 3 independent experiments. Error bars corresponding to the single-cycle infections were too large to include.

RSV infection of air-liquid-interface differentiated primary human airway epithelial (HAE) cells has been characterized to proceed through the apical surface of ciliated epithelial cells over a 3-to-4-day period in the presence of mucus production and without cytopathic effects (9). The treatment window of the different RSV inhibitors was assessed in a multicycle, 6-day infection of HAE cells. The addition of RSV inhibitors was delayed by 0, 24, or 48 h postinfection, all samples were harvested at 6 days postinfection, and virus replication was measured by RT-qPCR analysis (Fig. 3B). All compounds were nearly 100% effective when added immediately upon infection. The TMC-353121 fusion inhibitor showed declining activity with extended treatment delays, becoming 50% less effective when added 24 h postinfection. Similarly, RSV-604, despite being a postentry inhibitor, also lost activity as the infection progressed. RSV-604 was only 20% effective when added 48 h postinfection in HAE cells. The mechanism of this temporal loss of activity is not clear, but the effects are consistent with the decrease of activity observed over time in single-cycle (Fig. 3A) replication assays in A549 cells. Both the YM-53403 and BI-D direct-acting polymerase inhibitors were active regardless of the delay in addition of compounds.

Effect of antiviral compounds in the RSV replicon assay.

The inhibitors were next examined in a cell-based replicon assay. This assay uses BHK cells containing a stably transformed subgenomic RSV replicon expressing a GFP reporter cassette. The replicon is maintained by the expression of active RSV polymerase, and polymerase activity can be monitored by measuring GFP expression. Therefore, this assay is useful in identifying inhibitors of viral transcription and/or replication and measuring their potencies. The TMC-353121 fusion inhibitor was inactive in this assay. However, a postentry inhibitor, RSV-604, was also inactive, supporting the idea of the presence of a mechanism of action either upstream of or indirectly regulating viral transcription and/or replication. In contrast, bothYM-53403 and BI-D were active against the RSV replicon, with EC₅₀s of 0.3 and 0.4 μM, respectively (Table 5). The corresponding EC₅₀s corresponding to the antiviral activity of each compound in BHK cells infected with RSV-A2 virus were measured in the replicon cell line, and values were equivalent to those obtained with A549 or HEp-2 cells, with the exception of BI-D (Tables 1 and 5). The 5-fold-lower potency of BI-D in both infectivity and replicon assays in BHK cells may reflect the lack of an interferon response to RSV in this cell line (25, 26), which might be partly responsible for suppressing viral infection in BI-D-treated A549 and HEp-2 cells. These results confirm that, while RSV-604 either acts upstream of or indirectly regulates viral transcription and/or replication, YM-53403 and BI-D specifically target the RSV transcription and/or replication processes.

TABLE 5.

Activity of antiviral compounds in the RSV replicon assay^a

Compound	Infectivity (BHK EC50 [μM])	BHK replicon level (μM)
		EC50	CC50
TMC-353121	0.001 ± 0.0003	>0.5	>0.5
RSV-604	10.8 ± 3.6	>100	>100
YM-53403	0.2 ± 0.1	0.2 ± 0.1	>10
BI-D	0.2 ± 0.02	0.4 ± 0.02	2.3 ± 0.5

^aData were averaged from the results of a minimum of 3 independent experiments.

Effect of antiviral compounds on RSV RNA synthesis by isolated RNP complexes.

The RSV replicon assay identified YM-53403 and BI-D as compounds that target viral polymerase transcription and/or replication processes but did not explain the mechanism. To further define the YM-53403 and BI-D mechanisms, we performed assays using extracts containing RSV ribonucleic protein (RNP) biochemically enriched from RSV-infected HEp-2 cells. The RNPs can engage in RNA synthesis in vitro, and products are detected by incorporation of [³²P]CTP. Parallel reactions were performed with RSV and mock-infected cell extracts to control for incorporation of radiolabeled CTP into RNA by host cell polymerases. Since the majority of in vitro mRNA products generated in this assay contain poly(A) tails [i.e., are purified by oligo(dT); data not shown], the products that are detected are transcription products. Also, since the extracted RNP contains polymerase that is already associated with the template, the assay measures transcriptional elongation rather than initiation. Using this assay, RSV-604 and YM-53403 showed no significant effects on the amount or quality of RNA products (Fig. 4A, compare lanes 2, 3, and 4). In three assays, the signal of full-length transcripts was similar to that measured for controls and the observed changes in lower-molecular-weight species seen with YM-53403 were modest and poorly resolved. The EC₅₀s calculated by quantitative densitometry for RSV-604 and YM-53403 compounds were >50 μM. The BI-D compound had a significant effect, completely inhibiting full-length product formation and resulting in a smear of truncated RNAs (Fig. 4A, lane 5). The EC₅₀ calculated from the titration of the BI-D compound in the RNP reaction followed by densitometric quantitation of only full-length RNA products was 1 μM, and EC₅₀ calculated from integration of all RNA products was 12.3 μM (data not shown). These experiments showed that although they are both polymerase inhibitors, YM-53403 and BI-D have significantly different effects on the quality of RNA products generated by the RSV polymerase.

FIG 4.

Effects of postentry inhibitors on RNA synthesis in RNP or L/P biochemical assays. All compounds were tested at a fixed concentration of 50 μM. (A) In the RNP assay, RSV polymerase is extracted from infected HEp-2 cells in heterogeneous, preformed complexes, and the transcription of full-length mRNA products was measured. nt, no transcription. (B) RNA polymerase assay using purified complexes of L and P proteins coexpressed in baculovirus cells and an oligonucleotide template representing the promoter for genome synthesis (with the +3 and +1 start sites calibrated using ladders 1 and 2, respectively). Lanes 3 and 4 show the results of negative-control reactions in which the L/P complex was omitted and of negative-control reactions in which the L/P complex contained a substitution at amino acid N812 of the L protein, respectively. The data are representative of the results of three independent experiments performed with two different wild-type L/P preparations.

YM-53403 and BI-D elicit different patterns of RNA synthesis at the promoter.

The RSV polymerase can be reconstituted by the coexpression and purification of the RSV L and P proteins from baculovirus cells and RNA synthesized in vitro using an oligonucleotide template containing an RSV promoter sequence (21). A 25-nucleotide (nt) template corresponding to the 3′ end of the RSV trailer (antigenomic) promoter can be used to initiate RNA synthesis. This promoter has been shown to signal two initiation events at the +1 and +3 sites, with the +3 initiation site being dominant. The significance of the +3 initiation site is not yet known, but it yields a short uncapped transcript in RSV-infected cells that might have a functional role. This system was used to evaluate the ability of compounds to inhibit the expression of RSV polymerase during the initiation step of RNA synthesis, at a step after N protein displacement from the template. In the case of the DMSO control reaction, a series of bands could be detected, similarly to what has been observed previously (21). The bands that are less than 23 or 35 nucleotides in length presumably represent prematurely terminated products, and a dominant band detected at 21 nt has previously been shown to be RNA initiated at position +3 and then released 2 nt before the end of the template. The pattern of RNA synthesis in the presence of the RSV-604 compound was similar to that seen with the DMSO control (Fig. 4B, lanes 5 and 6), consistent with its putative mechanism of inhibition involving the N protein (which was absent in this assay). In contrast, YM-53403 inhibited accumulation of all RNA synthesis products (Fig. 4B, lane 7). This suggests that YM-53403 inhibits an early step of RNA synthesis at the promoter. Interestingly, inclusion of BI-D in the reaction resulted in increased accumulation of a 23-nt product and a reduction in accumulation of smaller products (Fig. 4B, lane 8). This suggests that BI-D results in hyperprocessivity of the polymerase. Our L/P assay was limited to the elongation of a 25-nt product, but we expect that the hyperprocessivity caused by BI-D in full-length replication might disrupt the kinetics of the elongation process and lead to early termination. Indeed, levels of full-length RSV poly(A) mRNA products in infected cells were dramatically reduced upon BI-D treatment (Fig. 1B). Taken together, the results in Fig. 4 show that, although YM-53403 and BI-D are both RSV polymerase inhibitors, they have very different effects on the quality of the RNA products that are generated. This likely explains why treatment of RSV-infected cells with these two compounds elicited opposing innate immune responses.

DISCUSSION

Our profiling of the TMC-353121 fusion inhibitor and three unique members of classes of published postentry inhibitors, RSV-604, YM-53403, and BI-D, led to new understandings of their antiviral mechanisms of action and interactions with the type I/III interferon responses in airway epithelial cells. While these compounds were all effective at inhibiting RSV replication in cell culture, they had differential effects on the host innate immune response to infection that could be explained by their distinct mechanisms of action.

TMC-353121 is a fusion inhibitor and blocks early steps in the virus entry process. This compound was most effective in reducing virus replication when added before or at the time of infection (Fig. 1 and 3), which is prior to an innate immune response. Our data provide evidence indicating that members of the class of fusion inhibitors attenuate parts of the interferon response pathways in a manner correlating with reductions in the viral load (Fig. 1 and Tables 2 and 3). Results of human experimental RSV challenge studies in healthy adult volunteers demonstrate that, in an upper respiratory tract model, viral load can drive disease in vivo (27). This outcome is further supported by the safety and efficacy of a potent RSV fusion inhibitor, GS-5806, currently in clinical development (20).

RSV-604 targets the N protein to inhibit RSV replication and acts at an early postentry step in the virus life cycle. The idea of this early life cycle mechanism of action is supported by the loss of its antiviral activity at later time points in the time of addition (Fig. 3) and in assays with a 24-h treatment delay (Fig. 2) as well as by its complete lack of activity in the replicon and RNP assays (Table 4 and Fig. 4). These data are consistent with results of previously published mechanism-of-action studies of this compound by Challa et al. (15). Challa et al. reported inactivity of RSV-604 in some cell lines (including BHK cells) infected with RSV and an inhibition of both RNA synthesis and the infectivity of virus progeny in HeLa cells. We observed equivalent levels of antiviral activity in all the cell lines that we used and no evidence of an effect of this compound on progeny virus formation in A549 or HEp-2 cells. It is possible that the RSV-604 compound structures (Table 1) and/or the BHK cell lines used in the study by Challa et al. were not identical to those used in ours. Our data suggest that RSV-604 inhibits RSV infection by as-yet-unknown early postentry events, indirectly affecting viral RNA transcription and replication. Like those of the fusion inhibitor, the effects of RSV-604 on RSV-induced innate immune signaling correlated with the antiviral activity of the compound.

YM-53403 and BI-D directly target the RSV RNA polymerase, are effective at inhibiting viral replication when added later in the infection cycle, and are also active in the RSV replicon assay. While both of these compounds target the polymerase, they have different mechanisms of action and opposite effects on RSV-induced innate immune signaling.

YM-53403 is a polymerase inhibitor with an as-yet-unknown mechanism. Our data determined using recombinant polymerase indicate that YM-53403 inhibits polymerase activity during an early stage of RNA synthesis (Fig. 4B). The relative inactivity of YM-53403 in the biochemical RNP assay may reflect the fact that polymerase already engaged in transcription is resistant to the compound or that the compound inhibits initiation specifically (Fig. 4). Importantly, the data we have obtained suggest that, in the context of RSV infection, YM-53403 suppresses overall RSV RNA synthesis but likely does not result in an increase in the levels of abortive RNAs. The location of YM-53403 resistance mutations in variable region 2 of the L gene, combined with structural domain studies of the RSV RNA polymerase indicating a dimerization function in this area (16, 28, 29), suggests that the compound may allosterically inhibit a tertiary complex essential for transcription or replication. Our findings are consistent with a recent publication showing the ability of a closely related compound, AZ-27, to inhibit a common step in the initiation of RSV transcription and replication (30). The attenuation of RIG-I-like pathways (Fig. 2 and Tables 2 and 3) is consistent with our finding that YM-53403 does not generate immunogenic RNA synthesis byproducts or interfere with a process recognized by cytoplasmic foreign RNA surveillance.

The BI-D compound is thought to interfere with the mRNA capping activity in domain V of the RSV polymerase, although the exact mechanism by which it does this is not known. Our data showing activity inhibition leading to abortive products in the RNP transcriptional assay are consistent with a capping defect (Fig. 4A). In the recombinant polymerase assay, BI-D caused hyperprocessivity of the polymerase (Fig. 4B). It is unclear whether the product being affected was RNA initiated at +1 (which would represent replication product) or RNA initiated at +3 (which would represent transcription product) or both. Further experiments are ongoing to tease apart the effects of BI-D on transcription and on replication. In either case, it is possible that in the context of infection, increased polymerase processivity could result in abortive transcripts that are likely responsible for the effect on antiviral signaling pathways. In the case of RNA replication, the replicative RNA must become encapsulated to be elongated to the end of the template. If BI-D were to increase replicase processivity, it is possible that the polymerase reaches an encapsulation checkpoint before the encapsulation process has had sufficient opportunity to be initiated, resulting in aborted replication production. Likewise, in the case of transcription, it is thought that capping of the transcript must occur to avoid abortive release. Increased processivity could result in the polymerase reaching this checkpoint before it has had an opportunity to cap the mRNA. In either event, in contrast to YM-53403, BI-D clearly had a very marked effect on the quality of RNAs generated by the RSV polymerase and this likely explains why BI-D had an effect on innate immune responses that was opposite that of YM-53403, causing a significant increase in activation of RIG-I pathways (Tables 2 and 3). The BI-D compound continues to be a valuable tool in RSV research aimed at studying the mechanisms and regulation of the RSV polymerase.

RSV causes severe disease in the respiratory tracts of infected individuals, and yet the virus replicates in ciliated epithelial cells without causing direct cytopathic effects. Thus, RSV-induced respiratory symptoms could be caused by the immune response at the site of infection. One of the first steps in this response is activation of the innate immunity.

Recently, Sun et al. (31) demonstrated that defective viral genomes generated during natural RSV infections are stimulatory of the immune system via type III interferon responses and functionally restrict viral replication, weight loss, and lung inflammation in vivo.

Activation of the innate immune response as a consequence of treatment by a mechanism similar to that ascribed to the BI-D compound could be beneficial to the host. The enhanced interferon response could play a significant role in helping to clear RSV infection. Importantly, we showed here that BI-D was 5-fold more potent in A459 cells, in which interferon pathways are intact, than in BHK cells, which are deficient in interferon (compare Tables 1 and 5). This augmentation of the interferon response could be very valuable in terms of treatment with a small-molecule inhibitor, as it could allow the inhibitor to be used at a lower dose and would reduce the possibility of resistance mutations arising.

On the other hand, dysregulation of innate immune responses, both in intensity and duration, could alter the immune responses associated with RSV infection. While the effects of BI-D on innate immune signaling have not been characterized in vivo, the altered innate immune signaling could have profound effects on the immune response to infection.

We would like to know the in vivo consequences of these antiviral mechanisms in an RSV-infected host; however, the compounds we compared are unfortunately not suitable for in vivo study. Their solubility, bioavailability, toxicity, pharmacokinetics (PK), and exposure in the lung either are poor or have not been studied, and not controlling these parameters would limit the usefulness of data. Further, small-animal models such as mouse and cotton rat RSV infections are not thought to be predictive of human disease and their innate immune responses to RSV are not established. Bovine or monkey would be most relevant, but as the compounds concerned are not being considered as possible therapeutics, cost and ethical considerations prohibit their study in these models. Thus, the observations presented here warrant careful testing of novel antiviral compounds to assess potential alterations in innate immune signaling pathways during RSV infection and to determine whether or not such alterations are beneficial or harmful. In vivo model systems to address the complexity of antiviral activity and immune responses in RSV are currently being investigated.